PySTARC - Python Simulation Toolkit for Association Rate Constants

PySTARC computes bimolecular association rate constants (k_on) via GPU-accelerated rigid-body Brownian dynamics.

Features

• GPU batch simulation - All trajectories run simultaneously on GPU via CuPy.
• Physics - Ermak-McCammon integrator, RPY hydrodynamics, Born desolvation, APBS electrostatics, adaptive time step, and Yukawa monopole fallback.
• Brownian bridge - Catches mid-step reaction crossings.
• Multi-GPU workflow - Split simulations across N GPUs with automatic grid generation, symlinked DX files, and pooled result combining.
• Automated system setup - From a PDB and topology file to a ready-to-run simulation in one command via setup.py.
• Convergence analysis - Wilson score CI, relative SE, and trajectory-count estimates for target precision.
• Output files - 14 structured files, including trajectories, encounters, radial density, angular maps, and transition matrices.
• Checkpointing - Automatic save/resume for long production runs.
• Live progress - k_on and P_rxn printed at configurable intervals.
• Temperature scaling - Correct thermodynamics at any temperature.

Installation

GPU (Linux/HPC):

git clone https://github.com/anandojha/PySTARC.git
cd PySTARC
bash install_PySTARC.sh

On Mac/CPU:

git clone https://github.com/anandojha/PySTARC.git
cd PySTARC
conda create -n PySTARC python=3.11 -y
conda activate PySTARC
conda install -c conda-forge ambertools apbs -y
pip install matplotlib pdb2pqr
pip install dist/pystarc-1.1.0-py3-none-any.whl --force-reinstall

Testing

python -m pytest tests/ -v          

Quick start

conda activate PySTARC
module load cuda                # HPC only, skip on local machines
cd examples/two_charged_spheres
chmod +x run.sh
bash run.sh

Examples

See examples/README.md for complete instructions.

License

MIT

Citation

When using PySTARC, please cite:

Ojha, A. A. et al. PySTARC: GPU-accelerated Brownian dynamics for bimolecular association rate constants (2026).

Requirements

• Python 3.11+
• AmberTools (tleap, cpptraj, ambpdb)
• APBS
• CuPy (GPU) or NumPy (CPU fallback)
• NVIDIA GPU with CUDA 12+ (recommended)